Preserve Failure Evidence Before Reworking a PCB Assembly

Reheating a suspect joint may restore operation while destroying evidence of why it failed. Cleaning can remove residues relevant to an intermittent leakage problem. Before changing a failed assembly, decide whether the immediate objective is recovery or failure analysis, and preserve enough evidence to support that decision.

Record the as-received condition

Identify the board, revision, population variant, serial number and reported use history. Photograph the assembly and the suspected region before handling alters them. Record visible damage, contamination, connectors, fasteners and any parts already removed. Separate observations from interpretations: a dark mark is an observation; overheating is a hypothesis.

Capture the original symptom as a reproducible test where possible. Include supply conditions, loads, software state, temperature and timing. If the failure is intermittent, record the sequence that preceded it and retain relevant logs. A statement such as ‘does not work’ leaves the investigation unable to distinguish several unrelated mechanisms.

Plan examinations in evidence-preserving order

Texas Instruments’ failure analysis guidance emphasizes useful failure history and appropriate handling when submitting devices for analysis. Component removal and handling can introduce damage that complicates interpretation. Discuss the planned examination with the laboratory when the suspected mechanism may be affected by preparation.

Start with suitable nondestructive observations and electrical checks. Define the applied conditions carefully; a diagnostic test that exceeds the device’s limits can create a new failure. Keep a record of every subsequent intervention so the final analysis can distinguish the received state from changes introduced during investigation.

Use comparison units thoughtfully

A working board of the same revision and population can help identify unexpected waveforms or resistance paths. It does not prove that every difference is causal. Manufacturing variation, firmware and operating history may differ. Record those differences before attributing the fault to the first unusual measurement.

If several failed units are available, preserve at least one in its original condition when practical. Different examination methods may compete for the same evidence. Cross-sectioning one location, for example, can prevent a later electrical test of that intact structure.

Authorize rework as a documented experiment

When rework is justified, state the hypothesis being tested and the observations expected if it is correct. Record materials, component identity and process conditions. Test again using the original reproduction sequence rather than only checking that the board powers up.

A successful repair narrows the investigation but does not always establish root cause. Distinguish confirmed findings, plausible mechanisms and unresolved questions in the report. That distinction prevents an effective temporary repair from becoming an unsupported claim about the cause of a field failure.